System overview

The mdx system consists of generic CPU nodes, GPU acceleration nodes, and storages. Resources are provided through virtual machines.

Compute Nodes

 Generic CPU NodesGPU Acceleration Nodes
ModelPRIMERGY CX2550 M6PRIMERGY GX2570 M6
CPU (per Node)Intel Xeon Platinum 8368 Processor
(38 core, 2.4 GHz) × 2
Intel Xeon Platinum 8368 Processor
(38 core, 2.4 GHz) × 2
Memory256 GiB512 GiB
GPU (per Node) NVIDIA Tesla A100 × 8
(GPU Memory 40 GiB)
Number of Nodes36840
Total Theoretical Computational Performance (Double Precision)2.1 PFLOPS6.4 PFLOPS
Total Theoretical Computational Performance (Single Precision) 6.7 PFLOPS
Total Theoretical Computational Performance (Half Precision) 100.7 PFLOPS

Storage

 ModelMain Applications
Virtual DiskIntelliFlash HD2160Storage area for virtual machine OS
High-Speed StorageDDN ES400NVX (SSD-based, 252 GByte/sec)/fast. Shared directory between virtual machines within a project.
For data requiring fast I/O, such as intermediate data during computation.
Large-Capacity StorageDDN ES7990X (SSD + HDD Hybrid, 157.5 GByte/sec)/large. Shared directory between virtual machines within a project.
For data requiring large storage capacity, such as large-scale training data.
Object StorageDDN ES7990X (SSD + HDD Hybrid, 63.0 GByte/sec)S3-compatible shared storage. For data sharing outside of mdx.

Virtual Machines

In mdx, you can create custom virtual machines to fit your specific needs. Virtual machines are built in units of CPU and GPU packs.

 CPU PackGPU Pack
CPU Cores1 Virtual Core18 Virtual Cores
GPUs 1 GPU
Memory1.51GB57.60GB
GPU Memory 40GB
Maximum # Packs Assignable to 1 Virtual Machine152 Packs8 Packs
Total Theoretical Computational Performance (Double Precision)Approx. 38.35 GFLOPSApprox. 20.2 TFLOPS
Total Theoretical Computational Performance (Single Precision) Approx. 20.9 TFLOPS
Total Theoretical Computational Performance (Half Precision) Approx. 315 TFLOPS